Effect of smoking on tuberculosis treatment outcomes: A systematic review and meta-analysis

• Published in: PloS one Vol. 15; no. 9; p. e0239333
• Main Authors: Abay Burusie, Fikre Enquesilassie, Adamu Addissie, Berhe Dessalegn, Tafesse Lamaro
• Format: Journal Article
• Language: English
• Published: Public Library of Science (PLoS) 01-01-2020
• ISSN: 1932-6203
• DOI: 10.1371/journal.pone.0239333

IntroductionNumerous studies have explored an effect of cigarette smoking on tuberculosis treatment outcomes but with dissimilar conclusions.ObjectiveTo determine the effect of cigarette smoking on tuberculosis treatment outcomes.MethodsPubMed, Cochrane library and Google scholar databases were searched last on February 27, 2019. We applied the random-effects model for the analysis. Publication bias was assessed using funnel plot and Egger's regression. Furthermore, we performed Orwin's Fail-Safe N and cumulative meta-analysis to check for small studies' effect.ResultsOut of 22 studies we included in the qualitative synthesis, 12 studies reported p-values less than 0.05 where smoking significantly favored poor treatment outcomes. The remaining 10 studies reported p-values larger than 0.05 implying that smoking does not affect the treatment outcomes. Twenty studies met the criteria for inclusion in a meta-analysis. The meta-analysis found that smoking significantly increased the likelihood of poor tuberculosis treatment outcomes by 51% (OR = 1.51; 95% CI = 1.30 to 1.75 and I-square = 75.1%). In a sub-group analysis, the effect was higher for low- and middle-income countries (OR = 1.74; 95% CI = 1.31 to 2.30) and upper-middle-income economies (OR = 1.52; 95% CI = 1.16 to 1.98) than for high-income ones (OR = 1.34; 95% CI = 1.03 to 1.75) even though the differences in the effects among the strata were not statistically significant as demonstrated by overlapping of confidence intervals of the effects. Meta-regression analysis, adjusted for income economies, found the effect of smoking has not significantly improved over the years (p = 0.92) and thus implying neither of the covariates were source of the heterogeneity. Egger's regression test indicated that publication bias is unlikely (p = 0.403).ConclusionCigarette smoking is significantly linked with poor tuberculosis treatment outcomes.